Transcoding as from MPEG-2 SDTV to MPEG-4 CIF reuses motion vectors and downsamples in the frequency (DCT) domain with differing treatments of frame-DCT and field-DCT blocks, and alternatively uses de-interlacing IDCT with respect to the row dimension plus deferred column downsampling for reference frame blocks.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A method of downsampling a two-dimensional block of discrete cosine transform (DCT) coefficients, comprising: (a) providing a two-dimensional N×N block of DCT coefficients; (b) applying a one-dimensional N/2×N de-interlacing inverse discrete cosine transform (IDCT) with respect to a first dimension of said block; and (c) applying a one-dimensional de-interlacing inverse discrete cosine transform (IDCT) with respect to a second dimension of the results of step (b); (d) wherein said de-interlacing IDCT is x e =T t (N/2) z p +QT t (N/2)K t z r , where x e is a vector of four downsample values, z p is an N/2 component vector of the even-index components of a column of coefficients in said first dimension of said block, said even-index components in bit-reversed order, Z r is an N/2 component vector of the odd-index components of said column of coefficients in said first dimension of said block, said odd-index components in bit-reversed order, T t (N/2) is the N/2-point IDCT, K=RLR t , where R is a bit-reversal permutation matrix; and L is a N/2×N/2 lower-triangular matrix, and Q is a N/2×N/2 diagonal matrix: diag[ cos((4m+1)/2N)] for m=0, 1, . . . , N/2 1.
2. The method of claim 1 , claim wherein: (a) said block is 8×8.
3. A method of downsampling a two-dimensional block of discrete cosine transform (DCT) coefficients, comprising: (a) providing a two-dimensional N×N block of DCT coefficients; (b) applying a one-dimensional N/2×N de-interlacing inverse discrete cosine transform (IDCT) with respect to a first dimension of said block; (c) applying a one-dimensional inverse discrete cosine transform (IDCT) with respect to a second dimension of the results of step (b); and (d) downsample the results of step (c) with respect to said second dimension; (e) wherein said de-interlacing IDCT is x e =T t (N/2) z p +QT t (N/2)K t z r , where x e is a vector of four downsample values, z p is an N/2 component vector of the even-index components of a column of coefficients in said first dimension of said block, said even-index components in bit-reversed order, z r is an N/2 component vector of the odd-index components of said column of coefficients in said first dimension of said block, said odd-index components in bit-reversed order, T t (N/2) is the N/2-point IDCT, K=RLR t , where R is a bit-reversal permutation matrix; and L is a N/2×N/2 lower-triangular matrix, and Q is a N/2×N/2 diagonal matrix: diag[cos((4m +1) /2N)]for m =0, 1, . . . , N/2 1.
4. The method of claim 3 , wherein: (a) said block is 8×8.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
September 17, 2003
March 11, 2008
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.